Noisy-LSTM: Improving Temporal Awareness for Video Semantic Segmentation

Semantic video segmentation is a key challenge for various applications. This paper presents new model named Noisy-LSTM, which trainable in an end-to-end manner, with convolutional LSTMs (ConvLSTMs) to leverage the temporal coherence frames, together simple yet effective training strategy that replaces frame given sequence noises. Our spoils frames and thus makes links ConvLSTMs unreliable; this may consequently improve ability of extract features from serve as regularizer avoid overfitting, without requiring extra data annotations or computational costs. Experimental results demonstrate proposed can achieve state-of-the-art performances on both CityScapes EndoVis2018 datasets. The code method available at https://github.com/wbw520/NoisyLSTM.